Maurice 7a40225852 feat(llm-parse): read bookings on a ChatGPT plan ...

A ChatGPT sign-in only ever ran the assistant: parsing's clients call an
endpoint with a key, and a plan has none. AI connections now lends the plan
as a reader. Each document is one Responses call with the schema as the one
tool the model has to call (tool_choice required). The token is read and
refreshed per call by ChatgptAuthService, and a sign-in that has to be
renewed lends nothing.

- ResolvedLlmConfig is a union: ApiLlmConfig (what an admin configures) or
  PlanLlmConfig (provider chatgpt, a read function, never stored).
- ConnectionReadClient sends the text plus photo or scanned pages, and reads
  the filled shape, or the shape written as text; prose is unreadable.
- AiChatClientService.chatgptPlan is the one place that builds the plan's
  client, for the assistant and for parsing.
- Wiki: a ChatGPT sign-in reads bookings on its plan.

2026-11-13 20:49:08 +01:00

..

booking-total.test.ts

fix(bookings): count a multi-segment rail ticket's fare once

2026-10-29 20:51:28 +01:00

clients.test.ts

fix(llm-parse): ask Anthropic models for the extraction tool instead of forcing it

2026-10-30 21:54:19 +01:00

connection-read.client.test.ts

feat(llm-parse): read bookings on a ChatGPT plan

2026-11-13 20:49:08 +01:00

currency-code.test.ts

style: format server and shared with prettier

2026-10-10 21:23:37 +02:00

extraction-router.test.ts

fix(llm-parse): the local router reports how its reading ended

2026-11-13 17:40:53 +01:00

image-input.test.ts

feat(assistant): models that read images see the live page

2026-10-22 21:34:15 +00:00

lenient-json.test.ts

style: format server and shared with prettier

2026-10-10 21:23:37 +02:00

llm-capabilities.controller.test.ts

style: format server and shared with prettier

2026-10-10 21:23:37 +02:00

llm-client.factory.test.ts

feat(llm-parse): read bookings on a ChatGPT plan

2026-11-13 20:49:08 +01:00

llm-config.resolver.test.ts

feat(llm-parse): read bookings on a ChatGPT plan

2026-11-13 20:49:08 +01:00

llm-config.test.ts

feat(assistant): one AI tile for assistant, parsing and MCP, a planner-aware chat

2026-10-22 21:30:26 +00:00

llm-local.controller.test.ts

fix(llm-parse): the local model list and pull send the saved api key to an ollama behind an auth proxy

2026-10-12 21:04:23 +02:00

llm-local.service.test.ts

fix(ai): connect and read bookings through any OpenAI-compatible own server

2026-10-26 10:20:46 +01:00

llm-parse.service.test.ts

fix(llm-parse): the local router reports how its reading ended

2026-11-13 17:40:53 +01:00

llm-prompt.test.ts

fix(ai-import): read a long document's end and charge a shared total once (#2652, #2651)

2026-10-14 13:24:27 +02:00

llm-request-error.test.ts

feat(llm-parse): a typed outcome for every AI reading instead of warning text

2026-11-13 17:40:29 +01:00

mail-read.service.test.ts

fix(mail-ingest): type errors and formatting after joining the flight work

2026-11-13 18:03:14 +01:00

mail-read.test.ts

feat(llm-parse): read a forwarded booking mail and name its trip in one request

2026-11-13 17:42:44 +01:00

mime-email.test.ts

style: format server and shared with prettier

2026-10-10 21:23:37 +02:00

mime-walker.test.ts

test: no non-null assertions or awaits in a loop in the new mail import tests

2026-11-13 19:31:14 +01:00

nuextract.test.ts

style: format server and shared with prettier

2026-10-10 21:23:37 +02:00

ollama-format.client.test.ts

test(ai): fix the model-not-found pattern, guard the fetch mocks, and read the booking in an ORM context

2026-10-26 10:20:47 +01:00

receipt-read.test.ts

style: format server and shared with prettier

2026-10-10 21:23:37 +02:00

text-extract.test.ts

style: format server and shared with prettier

2026-10-10 21:23:37 +02:00

text-window.test.ts

fix(ai-import): read a long document's end and charge a shared total once (#2652, #2651)

2026-10-14 13:24:27 +02:00